A universal brown dwarf desert formed between planets and stars

Giant planets and brown dwarfs play a crucial role in star and planet formation, as they are situated at the boundary between planets and stars with uncertain formation mechanisms. Previous observational searches for the formation boundary were hampered by the lack of large unified samples of wide-orbit giant planets and substellar companions. A combined analysis of radial velocity and astrometry mitigates this problem and has significantly enlarged the sample. Here we present a rigorous statistical analysis of the sample of 55 giant planets, brown dwarfs and low-mass stellar companions orbiting FGK stars. We quantitatively analyze the occurrence rates of brown dwarfs and identify a distinct brown dwarf desert at approximately $30,M_\mathrm{J}$, with no evidence of disappearance up to 20 au. Unlike previous studies that predicted a declining planet occurrence rate beyond the water-ice line, we identify a new population of giant planets and low-mass brown dwarfs in this region. The metallicity and eccentricity trends in our sample suggest that these are the consequences of two different formation scenarios. Our combined population synthesis model successfully accounts for the observed brown dwarf desert, supporting the dual formation hypothesis.

💡 Research Summary

The authors present a comprehensive demographic study of substellar companions—ranging from giant planets (≈5 MJ) through brown dwarfs (≈30 MJ) up to low‑mass stars (≈120 MJ)—orbiting FGK dwarfs at semi‑major axes of 2–20 AU. By combining high‑precision radial‑velocity (RV) measurements with astrometric data from Hipparcos and Gaia, they assembled a relatively unbiased sample of 55 companions around 54 stars drawn from four long‑term RV surveys (HARPS‑GTO, CPS, AAPS, Magellan).

To quantify detection limits, the team performed an extensive injection‑recovery campaign: for each of the 790 target stars they injected 75 000 synthetic signals spanning a wide range of masses, periods, eccentricities, and inclinations, then attempted to recover them using the same pipeline applied to the real data. Detection efficiency Q(m,a) was mapped on a four‑dimensional grid (mass, period, semi‑major axis, mass ratio) and reliability R was assessed on the subset of known companions, requiring recovered parameters to be within 30 % (or 0.1 for eccentricity, 30° for inclination) of the injected values. This rigorous approach yields a well‑characterized completeness function across the parameter space.

Occurrence rates were estimated using two complementary methods. First, a detection‑efficiency‑weighted kernel density estimator (wKDE) provided a non‑parametric view of the 2‑D distribution, revealing a clear paucity of objects near 30 MJ. Second, a hierarchical Bayesian framework was employed to fit parametric models to the occurrence‑rate density Γ(m,a) = d²N/(d ln m d ln a). The preferred model combines a log‑normal component (characterized by amplitude A, mean μ, width σ) with a power‑law tail (amplitude C, index α). Model comparison via Pareto‑smoothed importance sampling leave‑one‑out cross‑validation (PSIS‑LOO‑CV) shows that the log‑normal + power‑law model outperforms a simple flat model by a substantial ΔELPD.

The key scientific result is the identification of a “brown‑dwarf desert” centered at ≈30 MJ that persists out to at least 20 AU. The desert separates two distinct populations: (1) low‑mass companions (≤30 MJ) that preferentially orbit metal‑rich stars and exhibit low eccentricities, consistent with formation by core accretion; (2) higher‑mass companions (>30 MJ) whose occurrence shows little metallicity dependence but higher eccentricities, indicative of formation via gravitational instability or disk fragmentation. The authors further demonstrate that a population‑synthesis model incorporating both pathways reproduces the observed occurrence‑rate surface, including the depth and width of the desert.

In summary, by leveraging a large, homogeneous sample with well‑characterized detection completeness, the study provides robust statistical evidence for a universal brown‑dwarf desert and supports a dual‑formation scenario: core accretion dominates below ~30 MJ, while gravitational instability dominates above it. These findings refine our understanding of the planet–brown‑dwarf–star continuum, set new constraints for planet‑formation theories, and guide future surveys (e.g., direct imaging, astrometry) targeting the critical mass regime where the two formation channels intersect.